dcicutils 7.11.0.1b9__py3-none-any.whl → 7.12.0.1b4__py3-none-any.whl

dcicutils/misc_utils.py

@@ -7,11 +7,11 @@ import datetime
 import functools
 import hashlib
 import inspect
-import math
 import io
 import json
-import os
 import logging
+import math
+import os
 import pytz
 import re
 import rfc3986.validators
@@ -21,8 +21,8 @@ import warnings
 import webtest  # importing the library makes it easier to mock testing
 
 from collections import defaultdict
-from dateutil.parser import parse as dateutil_parse
 from datetime import datetime as datetime_type
+from dateutil.parser import parse as dateutil_parse
 from typing import Optional
 
 
@@ -192,11 +192,7 @@ class _VirtualAppHelper(webtest.TestApp):
     pass
 
 
-class AbstractVirtualApp:
-    pass
-
-
-class VirtualApp(AbstractVirtualApp):
+class VirtualApp:
     """
     Wrapper class for TestApp, to allow custom control over submitting Encoded requests,
     simulating a number of conditions, including permissions.
@@ -1315,6 +1311,11 @@ def file_contents(filename, binary=False):
         return fp.read()
 
 
+def json_file_contents(filename):
+    with io.open(filename, 'r') as fp:
+        return json.load(fp)
+
+
 def camel_case_to_snake_case(s, separator='_'):
     """
     Converts CamelCase to snake_case.
@@ -1357,25 +1358,6 @@ def capitalize1(s):
     return s[:1].upper() + s[1:]
 
 
-"""
-Python's UUID ignores all dashes, whereas Postgres is more strict
-http://www.postgresql.org/docs/9.2/static/datatype-uuid.html
-See also http://www.postgresql.org/docs/9.2/static/datatype-uuid.html
-And, anyway, this pattern is what our portals have been doing
-for quite a while, so it's the most stable choice for us now.
-"""
-
-uuid_re = re.compile(r'(?i)[{]?(?:[0-9a-f]{4}-?){8}[}]?')
-
-
-def is_uuid(instance):
-    """
-    Predicate returns true for any group of 32 hex characters with optional hyphens every four characters.
-    We insist on lowercase to make matching faster. See other notes on this design choice above.
-    """
-    return bool(uuid_re.match(instance))
-
-
 def string_list(s):
     """
     Turns a comma-separated list into an actual list, trimming whitespace and ignoring nulls.
@@ -2337,73 +2319,3 @@ def parse_in_radix(text: str, *, radix: int):
     except Exception:
         pass
     raise ValueError(f"Unable to parse: {text!r}")
-
-
-def pad_to(target_size: int, data: list, *, padding=None):
-    """
-    This will pad to a given target size, a list of a potentially different actual size, using given padding.
-    e.g., pad_to(3, [1, 2]) will return [1, 2, None]
-    """
-    actual_size = len(data)
-    if actual_size < target_size:
-        data = data + [padding] * (target_size - actual_size)
-    return data
-
-
-class JsonLinesReader:
-
-    def __init__(self, fp, padded=False, padding=None):
-        """
-        Given an fp (the conventional name for a "file pointer", the thing a call to io.open returns,
-        this creates an object that can be used to iterate across the lines in the JSON lines file
-        that the fp is reading from.
-
-        There are two possible formats that this will return.
-
-        For files that contain a series of dictionaries, such as:
-            {"something": 1, "else": "a"}
-            {"something": 2, "else": "b"}
-            ...etc
-        this will just return thos those dictionaries one-by-one when iterated over.
-
-        The same set of dictionaries will also be yielded by a file containing:
-            ["something", "else"]
-            [1, "a"]
-            [2, "b"]
-            ...etc
-        this will just return thos those dictionaries one-by-one when iterated over.
-
-        NOTES:
-
-        * In the second case, shorter lists on subsequent lines return only partial dictionaries.
-        * In the second case, longer lists on subsequent lines will quietly drop any extra elements.
-        """
-
-        self.fp = fp
-        self.padded: bool = padded
-        self.padding = padding
-        self.headers = None  # Might change after we see first line
-
-    def __iter__(self):
-        first_line = True
-        n_headers = 0
-        for raw_line in self.fp:
-            line = json.loads(raw_line)
-            if first_line:
-                first_line = False
-                if isinstance(line, list):
-                    self.headers = line
-                    n_headers = len(line)
-                    continue
-            # If length of line is more than we expect, ignore it. Let user put comments beyond our table
-            # But if length of line is less than we expect, extend the line with None
-            if self.headers:
-                if not isinstance(line, list):
-                    raise Exception("If the first line is a list, all lines must be.")
-                if self.padded and len(line) < n_headers:
-                    line = pad_to(n_headers, line, padding=self.padding)
-                yield dict(zip(self.headers, line))
-            elif isinstance(line, dict):
-                yield line
-            else:
-                raise Exception(f"If the first line is not a list, all lines must be dictionaries: {line!r}")

dcicutils/scripts/run_license_checker.py

@@ -0,0 +1,77 @@
+import argparse
+
+from dcicutils.command_utils import script_catch_errors, ScriptFailure
+from dcicutils.lang_utils import there_are, conjoined_list
+from dcicutils.license_utils import LicenseOptions, LicenseCheckerRegistry, LicenseChecker, LicenseCheckFailure
+from dcicutils.misc_utils import PRINT, get_error_message
+from typing import Optional, Type
+
+
+EPILOG = __doc__
+
+
+ALL_CHECKER_NAMES = LicenseCheckerRegistry.all_checker_names()
+NEWLINE = '\n'
+
+
+def main():
+
+    parser = argparse.ArgumentParser(
+        description="Runs a license checker",
+        epilog=EPILOG,
+        formatter_class=argparse.RawDescriptionHelpFormatter
+    )
+    parser.add_argument("name", type=str, default=None, nargs='?',
+                        help=f"The name of a checker to run. "
+                        + there_are(ALL_CHECKER_NAMES, kind='available checker',
+                                    show=True, joiner=conjoined_list, punctuate=True))
+    parser.add_argument("--brief", '-b', default=False, action="store_true",
+                        help="Requests brief output.")
+    parser.add_argument("--debug", '-q', default=False, action="store_true",
+                        help="Requests additional debugging output.")
+    parser.add_argument("--conda-prefix", "--conda_prefix", "--cp", default=LicenseOptions.CONDA_PREFIX,
+                        help=(f"Overrides the CONDA_PREFIX (default {LicenseOptions.CONDA_PREFIX!r})."))
+
+    args = parser.parse_args()
+
+    with script_catch_errors():
+        run_license_checker(name=args.name, verbose=not args.brief, debug=args.debug, conda_prefix=args.conda_prefix)
+
+
+def show_help_for_choosing_license_checker():
+    PRINT("")
+    PRINT(there_are(ALL_CHECKER_NAMES, kind='available checker', show=False, punctuation_mark=':'))
+    PRINT("")
+    wid = max(len(x) for x in ALL_CHECKER_NAMES) + 1
+    for checker_name in ALL_CHECKER_NAMES:
+        checker_class = LicenseCheckerRegistry.lookup_checker(checker_name)
+        checker_doc = (checker_class.__doc__ or '<missing doc>').strip(' \t\n\r')
+        PRINT(f"{(checker_name + ':').ljust(wid)} {checker_doc.split(NEWLINE)[0]}")
+    PRINT("")
+    PRINT("=" * 42, "NOTES & DISCLAIMERS", "=" * 42)
+    PRINT("Park Lab is a research laboratory in the Department of Biomedical Informatics at Harvard Medical School.")
+    PRINT("Park Lab checkers are intended for internal use and may not be suitable for other purposes.")
+    PRINT("External organizations must make their own independent choices about license acceptability.")
+    PRINT("Such choices can be integrated with this tool as follows:")
+    PRINT(" * Import LicenseChecker and LicenseCheckerRegistry from dcicutils.license_utils.")
+    PRINT(" * Make your own subclass of LicenseChecker, specifying a doc string and appropriate constraints.")
+    PRINT(" * Decorate your subclass with an appropriate call to LicenseCheckerRegistry.register_checker.")
+    PRINT("")
+
+
+def run_license_checker(name: Optional[str],
+                        verbose=LicenseOptions.VERBOSE,
+                        debug=LicenseOptions.DEBUG,
+                        conda_prefix=LicenseOptions.CONDA_PREFIX):
+    if name is None:
+        show_help_for_choosing_license_checker()
+    else:
+        try:
+            checker_class: Type[LicenseChecker] = LicenseCheckerRegistry.lookup_checker(name)
+        except Exception as e:
+            raise ScriptFailure(str(e))
+        try:
+            with LicenseOptions.selected_options(verbose=verbose, debug=debug, conda_prefix=conda_prefix):
+                checker_class.validate()
+        except LicenseCheckFailure as e:
+            raise ScriptFailure(get_error_message(e))

{dcicutils-7.11.0.1b9.dist-info → dcicutils-7.12.0.1b4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 7.11.0.1b9
+Version: 7.12.0.1b4
 Summary: Utility package for interacting with the 4DN Data Portal and other 4DN resources
 Home-page: https://github.com/4dn-dcic/utils
 License: MIT
@@ -25,11 +25,9 @@ Requires-Dist: PyYAML (>=5.1,<5.5)
 Requires-Dist: aws-requests-auth (>=0.4.2,<1)
 Requires-Dist: boto3 (>=1.17.39,<2.0.0)
 Requires-Dist: botocore (>=1.20.39,<2.0.0)
-Requires-Dist: chardet (>=5.2.0,<6.0.0)
 Requires-Dist: docker (>=4.4.4,<5.0.0)
 Requires-Dist: elasticsearch (==7.13.4)
 Requires-Dist: gitpython (>=3.1.2,<4.0.0)
-Requires-Dist: openpyxl (>=3.1.2,<4.0.0)
 Requires-Dist: opensearch-py (>=2.0.1,<3.0.0)
 Requires-Dist: pyOpenSSL (>=23.1.1,<24.0.0)
 Requires-Dist: pytz (>=2020.4)

{dcicutils-7.11.0.1b9.dist-info → dcicutils-7.12.0.1b4.dist-info}/RECORD

@@ -25,14 +25,14 @@ dcicutils/exceptions.py,sha256=4giQGtpak-omQv7BP6Ckeu91XK5fnDosC8gfdmN_ccA,9931
 dcicutils/ff_mocks.py,sha256=6RKS4eUiu_Wl8yP_8V0CaV75w4ZdWxdCuL1CVlnMrek,36918
 dcicutils/ff_utils.py,sha256=3-0AD1FVKmpI7qleMoIqBLPubBAgH5qBmQGVo6-gWnk,72419
 dcicutils/function_cache_decorator.py,sha256=XMyiEGODVr2WoAQ68vcoX_9_Xb9p8pZXdXl7keU8i2g,10026
-dcicutils/glacier_utils.py,sha256=x4zRGeSBS9c3LeurjR2gvEr_ipDTVpULvRFsIMfOVrs,33704
+dcicutils/glacier_utils.py,sha256=Q4CVXsZCbP-SoZIsZ5NMcawDfelOLzbQnIlQn-GdlTo,34149
 dcicutils/jh_utils.py,sha256=Gpsxb9XEzggF_-Eq3ukjKvTnuyb9V1SCSUXkXsES4Kg,11502
 dcicutils/kibana/dashboards.json,sha256=wHMB_mpJ8OaYhRRgvpZuihaB2lmSF64ADt_8hkBWgQg,16225
 dcicutils/kibana/readme.md,sha256=3KmHF9FH6A6xwYsNxRFLw27q0XzHYnjZOlYUnn3VkQQ,2164
 dcicutils/lang_utils.py,sha256=cVLRUGyYeSPJAq3z_RJjA6miajHrXoi6baxF8HzHmLc,27797
-dcicutils/license_utils.py,sha256=OhOfTXFivvb6Y3tiJAb1b9Is-OTpBfZjC18M-RvqBqk,40456
+dcicutils/license_utils.py,sha256=AJ7AwUb7YsXwrrncuS5bLwz3B0YYOHAqKwgf1JPLj6w,63798
 dcicutils/log_utils.py,sha256=7pWMc6vyrorUZQf-V-M3YC6zrPgNhuV_fzm9xqTPph0,10883
-dcicutils/misc_utils.py,sha256=XisEQGMkHI7k5RiK-k4yeG8Zw00H8b-v9o2Y7mZyKb8,94548
+dcicutils/misc_utils.py,sha256=d30xwLFW41FwZVDAEYulWwyZUcLEzmD-pxsMlKH3mF4,91148
 dcicutils/obfuscation_utils.py,sha256=fo2jOmDRC6xWpYX49u80bVNisqRRoPskFNX3ymFAmjw,5963
 dcicutils/opensearch_utils.py,sha256=V2exmFYW8Xl2_pGFixF4I2Cc549Opwe4PhFi5twC0M8,1017
 dcicutils/project_utils.py,sha256=qPdCaFmWUVBJw4rw342iUytwdQC0P-XKpK4mhyIulMM,31250
@@ -42,15 +42,15 @@ dcicutils/redis_tools.py,sha256=rqGtnVUjNjTlCdL1EMKuEhEMAgRJMiXZJkrKuX255QA,6509
 dcicutils/redis_utils.py,sha256=VJ-7g8pOZqR1ZCtdcjKz3-6as2DMUcs1b1zG6wSprH4,6462
 dcicutils/s3_utils.py,sha256=a9eU3Flh8Asc8xPWLGP16A6UQ_FVwhoFQNqm4ZYgSQ4,28852
 dcicutils/scripts/publish_to_pypi.py,sha256=qmWyjrg5bNQNfpNKFTZdyMXpRmrECnRV9VmNQddUPQA,13576
+dcicutils/scripts/run_license_checker.py,sha256=psv3c1Of7h4V4yvh93iyI2F3JFPzdzQakKdq97JThRw,3653
 dcicutils/secrets_utils.py,sha256=8dppXAsiHhJzI6NmOcvJV5ldvKkQZzh3Fl-cb8Wm7MI,19745
-dcicutils/sheet_utils.py,sha256=bnnefjeTUL4ES7gtqThISXJKeli1AIFryu4h7Dt9dxw,47040
 dcicutils/snapshot_utils.py,sha256=ymP7PXH6-yEiXAt75w0ldQFciGNqWBClNxC5gfX2FnY,22961
 dcicutils/ssl_certificate_utils.py,sha256=F0ifz_wnRRN9dfrfsz7aCp4UDLgHEY8LaK7PjnNvrAQ,9707
 dcicutils/task_utils.py,sha256=MF8ujmTD6-O2AC2gRGPHyGdUrVKgtr8epT5XU8WtNjk,8082
 dcicutils/trace_utils.py,sha256=g8kwV4ebEy5kXW6oOrEAUsurBcCROvwtZqz9fczsGRE,1769
 dcicutils/variant_utils.py,sha256=2H9azNx3xAj-MySg-uZ2SFqbWs4kZvf61JnK6b-h4Qw,4343
-dcicutils-7.11.0.1b9.dist-info/LICENSE.txt,sha256=t0_-jIjqxNnymZoNJe-OltRIuuF8qfhN0ATlHyrUJPk,1102
-dcicutils-7.11.0.1b9.dist-info/METADATA,sha256=MER7N-gDAB5nz6YT51jT7aIu8_rHT2x65FBF5x3DN70,3084
-dcicutils-7.11.0.1b9.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-7.11.0.1b9.dist-info/entry_points.txt,sha256=Z3vezbXsTpTIY4N2F33c5e-WDVQxgz_Vsk1oV_JBN7A,146
-dcicutils-7.11.0.1b9.dist-info/RECORD,,
+dcicutils-7.12.0.1b4.dist-info/LICENSE.txt,sha256=t0_-jIjqxNnymZoNJe-OltRIuuF8qfhN0ATlHyrUJPk,1102
+dcicutils-7.12.0.1b4.dist-info/METADATA,sha256=WBf2fEjWMlOtieSs4nq5zbiThbHAYzQliH7gmJ_0L04,3003
+dcicutils-7.12.0.1b4.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-7.12.0.1b4.dist-info/entry_points.txt,sha256=8wbw5csMIgBXhkwfgsgJeuFcoUc0WsucUxmOyml2aoA,209
+dcicutils-7.12.0.1b4.dist-info/RECORD,,

{dcicutils-7.11.0.1b9.dist-info → dcicutils-7.12.0.1b4.dist-info}/entry_points.txt

@@ -1,4 +1,5 @@
 [console_scripts]
 publish-to-pypi=dcicutils.scripts.publish_to_pypi:main
+run-license-checker=dcicutils.scripts.run_license_checker:main
 show-contributors=dcicutils.contribution_scripts:show_contributors_main